Journal of Science: Vol. 10 | Article 65 Tschinkel

The foraging tunnel system of the Namibian Desert , Baucaliotermes hainesi

Walter R. Tschinkel

Department of Biological Science, Florida State University, Tallahassee, FL 32303

Abstract

The harvester termite, Baucaliotermes hainesi (Fuller) (: Nasutitermitinae), is an endemic in southern Namibia, where it collects and eats dry grass. At the eastern, landward edge of the Namib Desert, the nests of these are sometimes visible above ground surface, and extend at least 60 cm below ground. The termites gain access to foraging areas through underground foraging tunnels that emanate from the nest. The looseness of the desert sand, combined with the hardness of the cemented sand tunnels allowed the use of a gasoline- powered blower and soft brushes to expose tunnels lying 5 to 15 cm below the surface. The tunnels form a complex system that radiates at least 10 to 15 m from the nest with cross- connections between major tunnels. At 50 to 75 cm intervals, the tunnels are connected to the surface by vertical risers that can be opened to gain foraging access to the surrounding area. Foraging termites rarely need to travel more than a meter on the ground surface. The tunnels swoop up and down forming high points at riser locations, and they have a complex architecture. In the center runs a smooth, raised walkway along which termites travel, and along the sides lie pockets that act as depots where foragers deposit grass pieces harvested from the surface. Presumably, these pieces are transported to the nest by a second group of termites. There are also several structures that seem to act as vertical highways to greater depths, possibly even to moist soil. A census of a single nest revealed about 45,000 termites, of which 71% were workers, 9% soldiers and 6% neotenic supplementary reproductives. The nest consisted of a hard outer “carapace” of cemented sand, with a central living space of smooth, sweeping arches and surfaces. A second species of termite, Promirotermes sp. nested in the outer carapace.

Key Words: foraging, harvester termite, nest construction, architecture, construction Correspondence: [email protected] Associate Editor: Robert Jeanne was editor of this paper Received: 14 February 2008, Accepted: 7 July 2008 Copyright : This is an open access paper. We use the Creative Commons Attribution 3.0 license that permits unrestricted use, provided that the paper is properly attributed. ISSN: 1536-2442 | Vol. 10, Number 65 Cite this paper as: Tschinkel WR. 2010. The foraging tunnel system of the Namibian Desert termite, Baucaliotermes hainesi. Journal of Insect Science 10:65 available online: insectscience.org/10.65

Journal of Insect Science | www.insectscience.org 1 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

Introduction mound and extending 25 to 30 m to the dead wood on which the termites were Termites can be roughly grouped into those feeding (Ratcliffe and Greaves 1940; Hill species that nest within their food, usually 1942; Greaves 1962). In C. lacteus, tunnels wood, and those that nest elsewhere and were more or less radial, with few cross must leave their nest in order to forage for connections, but with shafts to deeper soil. food. Of the latter type, nests may be In N. exitiosus, the radial tunnels were arboreal or subterranean, centrally located cross-connected. Hill (1925) noted or dispersed into small, connected units. subterranean passages with flattened Most termites shun the open air, and travel lumena thickly floored with “rejectamenta” to and from the foraging area by way of radiating outward from a nest of the subterranean tunnels or covered galleries. Australian Mastotermes darwiniensis, but Many species also cover the foraged he did not trace these passages far. A material with sheet galleries before dining. particularly thorough study is that of Darlington (1982), in which the Among ground-nesting termites, nests may underground foraging passages of be hidden below ground, or they may be Macrotermes michaelsoni were exposed conspicuous features of the landscape, such and quantified. as the mounds of the southern African species of Macrotermes or Trinervitermes. Many termites do not build mounds that Given a central nest, the need to forage for show above ground, but construct entirely food and an aversion toward open air, it is subterranean nests, with tunnels to the obvious that many termites must create surface. The African harvester termite, subterranean foraging tunnel systems. Such Hodotermes mossambicus, is well studied systems, however, have rarely been studied, because of occasional subterranean and are usually hardly mentioned (if at all) encounters during the digging of trenches in reviews of termite biology. Even an for construction (Hartwig 1963, 1965; authoritative treatment, such as Noirot’s Coaton and Sheasby 1975). These (1970) review of the nests of termites, gives encounters revealed that nests are located short shrift to how termites travel from their an average about 1.4 m below ground, but nests to their foraging areas. Typically, it is can be as shallow as a few cm or as deep as assumed that the termites travel in 6.7 m. Large passages connect these subterranean foraging tunnels (e.g. Sands subterranean nests to each other, and 1961), and indeed, the few existing studies smaller passages give the termites access to of subterranean foraging tunnels have the surface where they dump excavated soil revealed tunnel systems of remarkable size and forage for grass. Foraged grass is first and scale (Howse 1970; reviewed by Lee placed into small, superficial chambers for and Wood 1971). Most mound-building later transportion to the nests and species exit their nests through subterranean consumption. foraging tunnels that run a few cm below the surface. In some species, the tunnels are None of the reports on subterranean gallery short, and the termites travel some distance systems describe architectural details of the on the ground surface, but in others, the tunnels themselves or how they are tunnels may extend 25 to 30 m (or even 60 constructed. This paper reveals the intricate m) from the mound. For example, the and subtle architecture of the foraging Australian termites Coptotermes lacteus, C. tunnels of the Namibian harvester termite, brunneus, C. acinaciformis and Baucaliotermes hainesi (Fuller) Nasutitermes exitiosus constructed systems (Termitidae: Nasutitermitinae), and with 9 to 30 tunnels emanating from the describes how this complex system

Journal of Insect Science | www.insectscience.org 2 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

probably serves the foraging needs of the from year to year. termites. Like other harvester termites, B. hainesi foragers cut pieces of grass on the Tunnel excavation and mapping ground surface, and carry these back to Nests of B. hainesi were regularly visible at their nest. The range of this species is the surface as small mounds of cemented limited to southern Namibia and the material 10 to 15 cm high. All excavation northwestern Cape Province of South work was completed between October 22 Africa (Coaton and Sheasby 1973). and November 3, 2007. Tunnels were initially exposed by trenching around the Materials and Methods nest to locate tunnels, and excavated outward from there. The looseness of the The study site dry sand, combined with the relative The study site was located at latititude - hardness of the cemented sand tunnels 24.9702, longitude 15.9323 (according to facilitated exposure of the tunnels. The sand Google Earth) in the NamibRand Nature over the tunnels was loosened with a soft Reserve, a private reserve of about 180,000 hand broom, and the loosened sand was ha. The soil was red sand largely stabilized blown away with a gasoline-powered lawn by the grasses, uniplumis blower (Husqvarna Model 356 BTx) (Video (Licht) De Winter and S. giessii Kers 1, Video 2). This process produced a (: ), with circular, bare areas shallow trench 10-15 cm deep with the 5 to 15 m in diameter termed “fairy circles” mostly intact tunnels in the bottom. (van Rooyen et al. 2004), and abundant Branches and intersections were sometimes trails crossing it in multiple followed, but, for many branches, only the directions. The site sloped gently from initial few cm were exposed, leaving an about 1100 m elevation at the base of unknown but substantial fraction of the Jagkop mountain to about 940 m just short entire tunnel system unexposed. Tunnels of the Bushman Hills. Our two excavations that were in use were distinguished from were at approximately 1085 to 1090 m abandoned tunnels because the former elevation. This area has an arid climate remained intact upon excavation and/or where rainfall averages between 50 and 150 contained live termites mm per annum but is highly variable

Figure 1. Tunnels in current use by the termites could usually be recognized by the termites found within them. Here, nasute soldiers are defending a broken tunnel. High quality figures are available online. Journal of Insect Science | www.insectscience.org 3 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

when broken (Figure 1). Abandoned tunnels The dissection and collection took two tended to break, and were often filled with days. Termites from the mound and each sand. quarter of the nest from the top downwards were preserved separately. A sample of 100 In this manner, large parts of the foraging each of workers, soldiers and neotenic tunnel systems of two focal nests were supplementary reproductives (there were no exposed, one located at (lat, long) - mature alates in the nest) were killed by 24.96960, 15.93284 and the second at - freezing and air-dried for later 24.96981, 15.93403. The exposed tunnel determination of dry weight. systems were mapped by making a series of overlapping digital photographs (with a Counts were carried out in the laboratory at scale) from a uniform height (~ 1 m), like Florida State University. The alcohol was aerial photographs, and then combining drained off, and the total weight of (wet) these into a photomosaic. A scale map was termites from each nest portion was then made from each photomosaic. determined. Haphazard subsamples were then taken, weighed and the termites of Nest census each type counted. Multiplying these counts Before exposing the second tunnel system, by the factor = (total weight/sample weight) the nest was excavated for census. The nest gave estimates for each nest quarter, and the was carefully broken into pieces, beginning sum of these gave the total for the nest. at the top, and all live termites, as well as grass pieces, were collected by aspiration Results and preserved in alcohol for later counting. The brushing and blowing removed the

semi-aggregated sand overburden to expose tunnels whose walls retained their integrity because they were constructed of cemented

sand (Figure 2). The tunnels are thus not simply hollows excavated in the sand, but have walls reinforced with what can be seen

as “termite concrete” (which is of an unknown nature). Although the tunnels broke upon rough handling, with care,

sections could also be removed for closer inspection, transport and photography. Video 1. Click image to view video. Download video Tunnel architecture Tunnel architecture was complex. In cross section, most tunnels showed a raised

central portion with deep pockets along both sides (Figure 3). Sections of tunnels freed of loose sand were rarely simple

tubes, but showed many bulges and bumps on their undersides (Figure 4). Careful dissection of tunnels and bumps showed

that the raised, central portion was a smooth roadway that ran the length of all tunnels, probably serving as the main travel path for

Video 2. Click image to view video. Download video the termites in the tunnels. Along both sides of this roadway were pockets of varying

Journal of Insect Science | www.insectscience.org 4 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

Figure 2. Brushing with a soft hand broom and blowing the sand away exposed the foraging tunnels a few cm below the ground surface. These tunnels were constructed of cemented sand and retained their structure despite brushing and blowing. High quality figures are available online.

Figure 3. Cross sections of tunnels almost always showed a raised central portion, as seen in these two representative views. The raised central portion was the highway on which the termites traveled, while the pocket to the sides served as temporary depots for foraged grass pieces. High quality figures are available online.

Figure 4. Cleaned of all loose sand, tunnels always displayed a lumpy appearance, as in this representative section, viewed both from the side and from below. The vertical extension in the side view is a riser opening to the ground surface. High quality figures are available online. Journal of Insect Science | www.insectscience.org 5 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

depth and geometry (Figure 5). Many of great care during excavation to keep them these contained pieces of grass harvested intact. In most images, the former location from the surface by foragers, so it is of risers is seen as a double opening reasonable to presume that the pockets because two upward legs of the tunnel serve as temporary depots for harvested broke below their point of junction. The grass waiting to be transported nestward, distance between risers averaged 50 to 75 possibly by a different group of termites cm among tunnels (SD 14 to 40 cm), than the group that harvested the grass. suggesting that the termites rarely needed to (Figure 6 shows a view of the underside of travel more than one half to one meter on approximately 1 m of tunnel. the surface.

Termites in the tunnels could gain access to The tunnels did not run a uniform depth the surface through vertical risers, 5 to 15 below the surface, but swooped up and cm in height, that could be opened to the down between the risers, with the high surface (Figures 4 and 7). Riser openings points at the riser junctions and the low were usually closed during the day, as this points about midway between risers (Figure termite species forages mostly at night. 8). The internal runway therefore had “a Risers were very fragile, and it required roller coaster” or wave geometry. Measured

Figure 5. When the lumps on the tunnels were opened, they revealed lateral pockets that often contained grass pieces and served as temporary depots for grass foraged from the surface and awaiting transport to the nest. Figure 6. A longer section of tunnel viewed from below, showing the consistently lumpy structure. The branch in the top section is not a riser but a junction with another tunnel. Tunnels typically narrow briefly where they emit risers to the surface (not visible in this view from below). High quality figures are available online.

Journal of Insect Science | www.insectscience.org 6 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

from the riser-tunnel connection to the One section also contained a tunnel that lowest upper tunnel surface between risers, descended to greater depth. the tunnel dip averaged 5 to 8 cm among tunnels, with a standard deviation of 1.5 to Tunnels frequently intersected or branched, 2 cm. One dip was 21 cm, but the sometimes in rather complex significance of this large deviation was configurations. Cut-offs that shortened unclear. travel distance at more or less- perpendicular intersections were common Careful wetting of the upper surface of (Figure 10). Near the nests, tunnel exposed sections of tunnel allowed for intersections tended to form rectilinear removal of the tunnel roof to expose the grids (Figure 11). Occasionally, tunnels depot and riser structure of two crossed without joining, a termite version of approximately two-meter-long sections a fly-over. (Figure 9). The upper image shows the tunnel before removal of the roof, and the The tunnel systems lower, after. Depots can be seen along the Over the course of several days, large parts entire length on both sides and were most of the tunnel systems of the two focal nests likely to contain grass adjacent to risers. were exposed . The total length of tunnels

Figure 7. Two examples of risers that connect the tunnels to the surface. Risers were always associated with an upward swoop of the tunnel. The lower view also shows two junctions with other tunnels and one tunnel descending to greater depths. The risers were typically closed when the termites were not foraging. Figure 8. A section of tunnel almost a meter long, showing that risers were typically associated with the upward swoop of the tunnels. Most of the height of the risers has been broken off in this view. High quality figures are available online. Journal of Insect Science | www.insectscience.org 7 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

Figure 9. A section of about 3.4 m of tunnel (upper) from which the roof has been removed (lower), exposing the regular disposition of lateral depots and the central runway on which the termites probably travel. Grass was more commonly found in depots near risers, but these tunnels had been disconnected from the central nest for two days, so the distribution of forage may not be representative of normality. High quality figures are available online.

Figure 10. Perpendicular intersections often showed cut-offs that shortened the travel distance. Figure 11. Near the nests, foraging tunnels often showed rectilinear arrangements. High quality figures are available online.

Journal of Insect Science | www.insectscience.org 8 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

exposed was 76 m within an area of roughly (there was no evidence that the tunnels 170 m2 in the first excavation and 110 m of ended where we stopped excavating); (5) tunnel in an area of about 300 m2 in the the tunnels in the first excavation (Figure second excavation. Figures 12 and 13 show 14) connected two live and one abandoned an approximately 120o panorama of each of nest, suggesting that colonies of this termite these and reveal the scale of the termite may occupy more than one nest, and that enterprise. The exposed foraging tunnels lie nests are sometimes abandoned; (6) in the bottom of the trenches visible in the connected nests also suggests that the entire images. Maps created from the suitable habitat may be underlain by a photomosaics of these excavations are network of foraging tunnels. shown in Figures 14 and 15. These reveal several key features: (1) the tunnels tend Access to deeper soil generally to radiate outward from the nest; In the second excavation, two structures (2) the many unexcavated side-branches looked like small subsidiary nests located suggest that the area is actually underlain along the tunnel system. Dissection showed by a dense network of intersecting tunnels, them not to be nests, but rather large, with no area more than a meter or so from a vertical tunnels that seemed to descend to tunnel; (3) the frequent placement of risers deeper soil (Figure 16). The tunnel was not to the surface means that the foraging area excavated below about 0.5 m, but there was of the termites is more or less saturated with no sign that the tunnel direction changed. access points, and that the termites need The second nest mound that was excavated travel only short distances on the ground turned out to be an abandoned nest that was surface; (4) the tunnels probably extend being used as a vertical tunnel to deeper soil outward much farther than was excavated (Figure 17).

Figure 12. A 120o panoramic view of the first excavation. The exposed tunnels lie in the trenches, and Paul is indicating the location of the nest. The tunnels contacted a fairy circle in the center, and passed through one at the right. The total length of tunnels was about 76 meters, and the area enclosed by them about 110 m2. Figure 13. A 120o panoramic view of the second excavation. The nest was at the center of this excavation, but was removed for analysis before the tunnels were exposed. The white square at the upper left is a 1 m2 sampling device. High quality figures are available online. Journal of Insect Science | www.insectscience.org 9 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

Figure 14. A scale map of the first excavation, showing tunnel locations, nest location and the relationship to two fairy circles. Dotted lines indicate unexcavated branches and red dots show the locations of risers to the surface. This system connected two live and one dead colony. Figure 15. A scale map of the second excavation. Coding similar to Figure 16, with the addition of the blue dots showing where tunnels descended to greater depth. High quality figures are available online. Journal of Insect Science | www.insectscience.org 10 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

Figure 16. A structure in line with a foraging tunnel connecting a horizontal tunnel to one descending to greater depth. The termites almost certainly had tunnels at least deep enough to reach damp soil. High quality figures are available online.

Figure 17. What at first appeared to be a nest was shown by excavation and dissection to be an abandoned nest being used as a vertical tunnel to greater depth. Most of the former nest chambers had been filled with sand. Note that the termites have constructed a second, parallel tunnel to depth. High quality figures are available online.

Journal of Insect Science | www.insectscience.org 11 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

In this case, the termites had also immature (but very small immatures, and constructed a narrower tunnel to deep soil eggs were not counted), 32,300 (71%; 36 g, next to the abandoned nest. Most of the dry) were workers and 4,100 (9.1%; 3 g, chambers in this abandoned nest had been dry) were soldiers. In addition, there were filled with sand, and only the central core about 2,800 (6.2%; 5 g, dry) immature was being used as a vertical tunnel, along reproductives with wing buds. No primary with the purpose-built tunnel next to the reproductives and no “royal cell” were former nest. found, suggesting this may have been a subsidiary nest (or calie). Dissection and census of a nest Before exposing the second tunnel system, The termites were not evenly distributed the focal nest (Figure 18) was excavated for within the nest. The above-ground mound dissection and census of the contained contained very few termites. About 63% of termites. The nest was constructed of a hard the termites were found in the second and outer “carapace” of cemented sand and third quarters of the nest, that is, the center filled chambers, and an interior living space or core, with only about 3% in the top of sweeping surfaces and arches of a dark, quarter and 13% in the bottom quarter. smooth material (stercoral carton), with However, because it took two days to fairly constant spacing between surfaces dissect the nest, this distribution does not (Figure 19). necessarily represent the natural distribution. The nest was home to about 45,000 termites, of which about 6,000 (13%) were

Figure 18. The nest from excavation 2, removed from the soil and ready for dissection. The mound on top was all that was exposed above ground level. Approximately 50 cm lay below ground. Figure 19. The internal structure of the nest consisted of swooping arches and surfaces at a fairly constant distance apart and composed of a dark, smooth material (probably sand and termite excreta). Note the thick “carapace” surrounding the living space in the bottom view. High quality figures are available online. Journal of Insect Science | www.insectscience.org 12 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

A great deal of grass was found in the nest The scale of this endeavor is of similar (Figure 20), but these grass clippings were magnitude as several other species of not evenly distributed. The top quartile (0- mound-building termites, including C. 10 cm) contained 3.6 g of grass, the next lacteus, C. brunneus, C. acinaciformis, N. 10.5 g, the third 0.75 g and the bottom exitiosus (Ratcliffe and Greaves 1940; Hill almost none. This distribution is probably 1942; Greaves 1962) and M. michaelsoni the result of the depth at which the nest (Darlington 1982). Lee and Wood (1971) connects to the foraging tunnels, about 10- suggest that the underground foraging 15 cm below ground, combined with the networks of subterranean termites are consumption of the grass as it is moved probably of great ecological importance. No deeper toward the core of the nest where the one who has been to Africa or Australia bulk of the termites were located. The total could argue with that claim. dry weight of grass in the nest was about 15 g, probably a small fraction of what was Previous reports on subterranean foraging still in the tunnel system depots. tunnels gave few architectural details of their construction. The only two exceptions Promirotermes sp. are Greaves (1962), who reported that the A species of smaller termite, tunnels of C. acinaciformis, a wood-feeding Promirotermes sp., was found co-nesting species, were made of cemented soil with a with B. hainesi. Several chambers simple, flattened lumen in which the containing workers, soldiers and termites traveled, and Darlington (1982), reproductives were located in the carapace who described part of the foraging passage surrounding the main B. hainesi nest. The system of the fungus-gardening M. relationship of this species to B. hainesi, the michaelsoni in great quantitative detail. The “host,” is unknown. aeolian sands of the Namib Desert were ideal for exposing architectural details Discussion because the surrounding sand could be loosened with a soft brush and blown away, Like many other species of termites, B. but the cemented sand that formed the hainesi operates on an impressive scale. tunnels remained intact, revealing subtle, Workers from each nest travel to and fro in complex and functional architecture. Such the foraging tunnel system, harvesting grass discrimination would have been difficult in from at least several hundred square meters. more compacted or fine-grained soils.

Figure 20. Abundant grass pieces were found in the second quarter of the nest, but the third quarter contained much less and the bottom almost none. Grass pieces entered the nest through connections with foraging tunnels about 10-15 cm below ground. High quality figures are available online. Journal of Insect Science | www.insectscience.org 13 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

Ironically, one of the clearest recent brigade,” a system of greater efficiency exposures of a subterranean termite tunnel than one in which each individual harvests system involved fossil termite nests dating and transports each piece of grass all the to the upper Miocene and Pliocene eras (3-7 way to the nest. Leafcutter ants also use million year ago) in Chad (Duringer et al. caching and “bucket-brigade” transport for 2007). These fossils were attributed to an leaf pieces (Hart and Ratnieks 2001; ancestral fungus gardening Macro- Anderson et al. 2002), thus partitioning the termitinae, and they consisted of many task of foraging into cutting, caching and small globular nests connected by multiple transporting stages. Caching was rectilinear side tunnels to a straight main more likely when traffic was heavy or tunnel up to tens of meters long. The entire bottlenecked, and incurred the cost of network of tunnels and chambers was all in mismatching the leaf piece with the size of a plane, with no evidence of vertical the subsequent transporting worker, thus connections. In this regard, the layout slowing transport. Anderson et al. (2002) seems somewhat similar to the nest arenas used simulations to test for optimality in of the fungus gardening Odontotermes such transport systems. It is likely that B. fulleri in which all chambers were located hainesi also tends to cache more grass less than 30 cm below the surface pieces when cutting rate exceeds transport. (Darlington 2007). The partitioning of foraging in this manner unlinks harvesting, a mostly nocturnal task Depots for foraged grass have been reported which carries the risk of exposure to for another harvester termite, the desiccation and predation, from transport, widespread Hodotermes mossambicus, but which is relatively safe within the tunnel the depots were small chambers around the system and can probably proceed more or nest perimeter or small chambers near the less around the clock, as it does in M. surface, rather than being part of the michaelsoni. The obvious advantage of foraging tunnels (Hartwig 1963, 1965; such a system may underlie the reason it Coaton and Sheasby 1975). Probably, this has evolved in such diverse taxa as ants and cache system evolved independently, as termites (and humans). these species belong to different subfamilies and the depots have different The results leave the spatial extent and size structures. However, Darlington (1982) of the colony of this termite undetermined. described and quantified depots along the We found no primary reproductives in the foraging tunnels of M. michaelsoni and the dissected nest and no structure that might Brazilian Syntermes molestus. The depots be a “royal cell.” Combined with the fact of M. michaelsoni were especially similar that at least two live nests only about 6 to those of B. hainesi, and Darlington meters apart were connected with tunnels speculates that termites foraging on suggests that a colony may consist of dispersed food such as grass or litter ought multiple nests, some possibly deep in the to evolve tunnel system with caches ground, as suggested by the existence of because foraging must occur in episodes. tunnels-to-depth. Fuller (1915, as cited in She calculates that the volume of caches Lee and Wood 1971) reported that adjacent underlying an area was similar to the mounds of Trinervitermes trinervoides were volume of forage gathered in that area in interconnected through subterranean one night. The presence of caches of grass tunnels. On the other hand, Darlington pieces in the depots of B. hainesi strongly (1982) found the remains of dead soldiers suggests that the workers that harvest the and workers in the contact zone between grass on the surface are distinct from the foraging tunnel systems of neighboring tunnel transport workers, and that the mounds of M. michaelsoni, suggesting the system is to some degree a “bucket- occurrence of territorial battles.

Journal of Insect Science | www.insectscience.org 14 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

conditions conducive to the growth of Ebeling and Pence (1957) described how grasses may occur less than annually, and Reticulitermes hesperus use fine soil then only for short periods. Estimates for particles mixed with saliva to line their grass consumption in a “saturated” tunnels. In light of the extreme aridity of population of H. mossambicus in a more the Namib Desert, and the fact that nest and lush habitat (Zululand) ranged up to 1 to 3 tunnel construction require water, it seems metric tons per ha, practically the total yield inescapable that the termites have access to of hay, but other estimates were much moist soil, probably at great depth. When lower (Coaton and Sheasby 1975). There first brought to the surface and dumped, soil are many reports of H. mossambicus excavated by H. mossambicus in the study creating bare spots through their harvesting area was damp (personal observation), yet activity. It has been suggested that this no trace of dampness was detectable even termite is the cause of the fairy circles in excavations over 2 m deep. Yakushev mentioned in the Materials and Methods (1968, as cited in Lee and Woods 1971) (Becker 2007), but this claim is contested reports that some termite species may make (van Rooyen et al. 2004). Darlington (1982) tunnels to moisture as deep as 70 m. estimated the nightly forage collected by M. michealsoni to be approximately 0.6 to 1.1 Photographs included in Hill’s (1942) kg. Finally, Darlington (1982) showed that treatise of Australian termites show that the surface access points in M. michaelsoni Coptotermes acinicaformis and C. lacteus, tunnels to be dense enough that termites both mound-builders, construct nests with a need rarely travel more than 10 cm from an very thick “carapace,” much like B. hainesi. opening to forage. The actual density of This feature is lacking in the other species access points in B. hainesi is unknown, but examined in Hill’s book. In contrast to the is clearly higher than indicated in Figures nests of B. hainesi, the nests of 14 and 15, because many of the cross- subterranean-nesting termites are often connecting passages were left unexcavated. surrounded by an empty space rather than a Likewise, Darlington (1982) estimated that “carapace” (Noirot 1970). Perhaps the the nest of M. michaelsoni has a total of 6 difference lies in the relative instability of km of permanent foraging tunnels, but in the dry sands in which B. hainesi nests. view of unexcavated cross-passages and the difficulty of placing colony boundaries on This estimate of the nest population is B. hainesi, a corresponding estimate is surely an underestimate of the actual undetermined for this study. B. hainesi population, for it is likely that a substantial colonies are much smaller than those of M. fraction of the termites were in the foraging michealsoni, yet their work is still tunnels at the time of collection. Even after impressive. It is likely that similar tunnel- removal of the nest, abundant termites were and-depot systems are characteristic of found in the tunnels during several days of many harvesting termites. excavation. Whether their home was in the collected nest or in another, possibly a Acknowledgements deeper nest, could not be determined. We are grateful to the NamibRand Nature Considering the density of foraging access Reserve for allowing us to do this research points as well as the biomass of termites in the reserve. It was heaven on earth. We and the amount of grass pieces found in the are particularly grateful to Danica Shaw and nest and tunnels, it is likely that B. hainesi Nils Odendaal for their generous and foraging has a considerable impact on the indispensable help in making arrangements, sparse grasslands of the eastern Namib finding accommodations and generally Desert. This is more likely because getting us started and keeping us going. I

Journal of Insect Science | www.insectscience.org 15 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

am greatly indebted to Vivienne Uys of the Palaeogeography, Palaeoclimatology, South African Biosystematics Division, Palaeoecology 251: 323–353. Protection Research Institute. She not only identified termites for me, but Ebeling W, and Pence RJ. 1957. Relation of provided me with several hard-to-find particle size to the penetration of termite references. We thank Paul Komagab subterranean termites through barriers of for helping with the daily chores of the sand or cinders. Journal otf Economic research and excavations. We gained Entomology 50: 690-692. invaluable insights into the ecology and history of the area from Jürgen and Dorothe Fuller C. 1915. Observations on some Klein, and from Albi Brückner. South African termites. Annals of the Natal Museum 3: 329-595 (as cited in Lee and References Wood 1971).

Anderson C, Boomsma JJ, Bartholdi JJ. Greaves T. 1962. Studies of the foraging (2002). Task partitioning in insect societies: galleries and the invasion of living trees by bucket brigades. Insectes Sociaux 49: 171- Coptotermes acinaciformis and C. brunneus 180. (Isoptera). Australian Journal of Zoology 10: 630-651. Becker T. 2007. Das Phänomen der Feenkreise (Fairy Circles) im Kaokoland Hart AG, and Ratnieks FLW. 2001. Leaf (NW Namibia) Basic and Applied Dryland caching in the leafcutting ant Atta Research 1: 121-137. colombica: organizational shift, task partitioning and making the best of a bad Coaton WGH, Sheasby JL. 1975. National job. Animal Behavior 62: 227-234. survey of the isopteran of southern Africa 10. The genus Hodotermes Hagen Hartwig EK. 1963. Termite (Isoptera) en (Hodotermitidae). Cimbebasia 3: 106-137. grondverplasing. In: Proceedings of the first National Symposium on Soil Biology. pp. Darlington JPEC 1982. The underground 240-251. Potchefstroom, Dec. 1963. (as passages and storage pits used in foraging cited in Coaton and Sheasby 1975). by a nest of the termite Macrotermes michaelsoni in Kajiado, Kenya. Journal of Hartwig EK. 1965. Die nestelsel van die Zoology, London 198: 237-247. grasdraertermiet Hodotermes mossambicus (Hagen)(Isoptera) en aspekte rakende Darlington JPEC 1993. Underground bestriding. South African Journal of foraging passages and storage pits built by Agricultural Science 8: 643-660 (as cited in the termite Syntermes molestus in Goiania, Coaton and Sheasby 1975). Brazil (Isoptera: Termitidae). Sociobiology 23: 211-212. Hill GF. 1925. Notes on Mastotermes darwiniensis. Proceedings of the Royal Darlington JPEC. 2007. Arena nests built Society of Victori 37: 119-124. by termites in the Masai Mara, Kenya. Journal of East African Natural History. Hill GF. 1942. Termites (Isoptera) from the 96: 73-81. Australian region. Australian Council for Scientific and Industrial Research, Duringer P, Schuster M, Genise JF, Melbourne. Mackaye HT, Vignaud P, Brunet M. 2007. New termite trace fossils: Galleries, nests Howse PE. 1970. Termites: a study in and fungus combs from the Chad basin of social behavior. Hutchinson University Africa (Upper Miocene–Lower Pliocene). Library.

Journal of Insect Science | www.insectscience.org 16 Journal of Insect Science: Vol. 10 | Article 65 Tschinkel

Lee KE, Wood TG. 1971. Termites and soils. Academic Press.

Noirot CH. 1970. The nests of termites, in Biology of Termites, vol. 2, Krishna K. and Weesner FM, eds. Academic Press.

Ratcliffe FN, Greaves T. 1940. The subterranean foraging galleries of Coptotermes lacteus (Frogg) Journal of the Council of Scientific and Industrial Research of Australia 13: 150-161 (as cited in Lee and Wood 1971).

Sands WA. 1961. Foraging behaviour and feeding habits in five species of Trinervitermes in West Africa. Entomologia Experimentalis et Applicata 4: 277-288.

van Rooyen MW, Theron GK, van Rooyen N, Jankowitz WJ, Matthews WS. 2004. Mysterious circles in the Namib Desert: review of hypotheses on their origin. Journal of Arid Environments 57: 467-485

Yakushev E. 1968. Influence of termite activity on the development of laterite soil. Soviet Soil Science 1968: 109-111 (as cited in Lee and Woods 1971).

Journal of Insect Science | www.insectscience.org 17